Conventionally, a storage apparatus is provided with a plurality of controllers and processing on accesses from a host computer is distributed among the plurality of controllers (for example, Patent Literature 1). If a failure occurs at one of the controllers in such a storage apparatus, the other controller where the failure has not occurred will process accesses from a host computer collectively. When this happens, the controller where the failure has occurred executes processing for identifying a failure location and notifies a user of the failure location.